Performance Analysis of Licensed-Assisted Access with Listen-Before-Talk in Non-Saturated Condition

Abstract

This paper analyzes the performance of licensedassistedaccess (LAA) systems conditioned on non-saturatedtraffic in unlicensed spectrum. Under the condition, we proposeMarkov chain models which describe time-domain behavior of aLAA system with different listen-before-talk (LBT) procedures.For evaluation of coexistence performance, a Wi-Fi access pointmodel is adopted and combined with the proposed models.We find an optimal contention window (CW) size of the LAAsystems in which total throughput of both networks is maximizedwhile satisfying the required throughput of each network, underthe given traffic densities of both networks. Total throughputimprovement is compared according to different LBT procedures.Numerical results show that the LAA systems with the optimalCW size lead to higher throughput improvement than withoutthe optimal CW in the coexistence scenario.ystem with different listen-before-talk (LBT) procedures.For evaluation of coexistence performance, a Wi-Fi access pointmodel is adopted and combined with the proposed models.We find an optimal contention window (CW) size of the LAAsystems in which total throughput of both networks is maximizedwhile satisfying the required throughput of each network, underthe given traffic densities of both networks. Total throughputimprovement is compared according to different LBT procedures.Numerical results show that the LAA systems with the optimalCW size lead to higher throughput improvement than withoutthe optimal CW in the coexistence scenario.